Know who you're calling before you call them.

Vertiqa's Prospect Research dispatches from an existing Account or a website URL, scores the prospect against your ICP using your kernel's sales config, cites every claim, and flags what couldn't be verified — proposed Account enrichment, cohort Tags, and an Account Document stay pending until you approve.

Book a demo See it in the app →

Shipped in v1.37 Sales Loop Agents Phase A

The spine of every Vertiqa agent

Every agent runs inside the same system of record — bounded by what it remembers, what it is allowed to do, and what gets written down.

Memory

Reads your sales config, the existing Account, and prior research

ICP, positioning, offer, claim guardrails, and cohort defaults from /settings/sales are snapshotted onto every run; the agent reads the existing Account if dispatched from one, or creates a draft Account in a research-pending state if dispatched from a URL.

Approval

The agent proposes; you apply

Research output is a set of proposed mutations — Account enrichment fields, cohort Tags, Activity entries, and an Account Document — alongside pending-verification flags. Nothing touches your CRM until you approve the dossier in the Approval Inbox.

Audit

Every claim has a citation or a pending flag

ICP Fit is computed deterministically from evidence; each claim carries citations; anything that couldn't be verified is flagged pending rather than guessed; the dossier is stored with the sales-config snapshot version so you can see exactly which guardrails were in force when the research ran.

What it does

Four concrete behaviors, every one traceable to a shipped v1.37 Sales Loop Agents Phase A requirement — no roadmap features, no aspirational claims.

Researches an account from a name or a website URL

Vertiqa's Prospect Research dispatches from an existing Account, or from a plain website URL when no Account exists yet. URL-start dispatch creates a draft Account in a research-pending state through the Kernel before the run begins, so every research dossier has a real Account record to attach to — and so the rest of your org can see that research is in flight.

Scores ICP fit deterministically from your sales config

ICP, positioning, offer, claim guardrails, and cohort defaults live in owner-editable sales settings at /settings/sales — seeded from your active pack so the agent speaks the right vocabulary on day one. ICP Fit is computed deterministically from the evidence the agent could verify against that config, with a documented fallback to pack and system defaults when your config has not yet been customized.

Cites every claim. Flags what it couldn't verify

Each finding in the research dossier carries citations to the sources it came from. Anything the agent couldn't verify is flagged pending rather than guessed, so reviewers see exactly where evidence is firm and where a human still needs to confirm. ICP Fit, Outreach Angle, Pre-send Disqualifier Checklist, reviewer markdown, and the proposed mutations all sit alongside the citations on one screen.

Proposes the mutations. You apply them

The research dossier is a set of proposed mutations, not a fait accompli. Approval is the moment Account enrichment fields, cohort Tags, Activity entries, timeline and audit evidence, and an Account Document are atomically written to the Kernel — all in one transaction, all through the same services your team already uses, and never piecemeal.

What this agent will not do

Never contacts the prospect. Reaching out is a separate, downstream step — Outreach Writer is the agent that handles outbound communications, and even that stays approval-gated. Prospect Research only researches.
Won't enrich an account without your approval. Proposed Account fields, cohort Tags, Activity entries, and the Account Document stay pending until you approve them in the Approval Inbox — approval is the moment they're atomically written to the Kernel.
Flags pending verification rather than guessing. Every claim carries a citation or a pending-verification flag; ICP Fit is computed deterministically from the evidence the agent could actually verify against your sales config, never fabricated.
Confined to the ICP and claim guardrails you configure in sales settings. Positioning, offer, claim guardrails, and cohort defaults come from /settings/sales — the agent doesn't learn off-policy claims and doesn't override your guardrails.

Works with the kernels you already use

Prospect Research reads and writes through the same Kernel surfaces your team uses — no separate vendor system, no shadow data store.

Sales config at /settings/sales
Owner-editable ICP, positioning, offer, claim guardrails, and cohort defaults; seeded from your active pack (HVAC, SI, Insurance, horizontal); snapshotted onto every research dossier so you can see which guardrails were in force at run time (L279, CFG-01..03).
Account record + draft accounts
Research dispatches from an existing Account or a website URL; URL-start creates a draft Account in a research-pending state through the Kernel before the run begins (L274, PR-02).
Approval Inbox
Proposed mutations land here with citations, pending-verification flags, and reviewer markdown; approval atomically applies Account enrichment, cohort Tags, Activity entries, timeline/audit evidence, and the Account Document (L276, PR-04).
Outreach Writer (downstream)
Once a research dossier is approved, Outreach Writer can produce personalized outbound from it. Outreach Writer refuses missing, pending, or rejected research — Prospect Research is the gate that makes downstream outbound possible. See /agents/outreach-writer.
Admin runs inspection
Provider failures, timeouts, rate limits, and runtime errors are visible in Admin UI; no run stays stuck in queued — failures surface as failed with an error message you can act on (L278, PR-06).
Pack vocabulary (HVAC / SI / Insurance / horizontal)
Sales config seeds from the active pack so ICP, positioning, and cohort defaults speak the right language for your industry — no raw JSON editing required.

Frequently asked

Where is my data stored?

Inside the same Vertiqa Kernel that powers every other agent and surface in your org. The research dossier, the proposed Account mutations, the citations, and the pending-verification flags all live in your organization's Kernel — not in a separate vendor system. Tenant isolation, org scoping, and permission gates are enforced at the endpoint level for every research run and every approval action.

What gets stored when a research run completes?

A research artifact tagged with the active sales-config snapshot version (stored as a `sales_loop_research_v1` artifact), the ICP Fit score with the evidence behind it, the Outreach Angle, the Pre-send Disqualifier Checklist, citations for every claim, pending-verification flags for anything the agent couldn't verify, reviewer markdown, and the exact proposed mutations — Account enrichment fields, cohort Tags, Activity entries, and an Account Document. Nothing is applied until you approve.

What happens if I reject the research?

Reject is a first-class outcome. No Account enrichment is applied, no cohort Tags are written, no Activity entries are created, and no Account Document is attached. The research dossier stays on record so the next run can see prior decisions and avoid re-suggesting what you already rejected, but nothing in your CRM changes until an approval action is run by a human.

Can I edit the research before applying it?

Yes. Reviewer markdown is part of the research dossier, and the Approval Inbox renders the proposed mutations alongside the citations and pending-verification flags so you can correct the dossier before approval. Approval is the moment Account enrichment, cohort Tags, Activity entries, timeline/audit evidence, and the Account Document are atomically written through Kernel services — never before, and never in a piecemeal way.

Which verticals is this built for?

Built first for relationship-driven service businesses where you need to know the prospect before you call. The pack vocabulary, the sales-config snapshot, and the ICP scoring fit life-insurance practices and travel advisors particularly well — see /life-insurance and /travel-advisors for vertical context. ICP Fit falls back to pack and system defaults when your sales config has not been customized yet, so the agent has a sensible starting point on day one.

Built for businesses that need to qualify before they call

Prospect Research ships with pack vocabulary, sales-config snapshots, and ICP scoring tuned for relationship-driven verticals where context wins the conversation.

Built for life-insurance practices that need to qualify before they call Built for travel advisors who need context before every consultation

Know who you're calling before you call them.

Book a demo to see Prospect Research turn a website URL into a citation-backed dossier — ICP Fit, Outreach Angle, pending-verification flags, and a set of proposed Account mutations you approve in one click.

Book a demo